Spatial Clustering in Slotted ALOHA Two-Hop Random Access for Machine Type Communication Ziwen Zhao, Sebastian S. Szyszkowicz, Tamer Beitelmal, Halim Yanikomeroglu Carleton University Ottawa, Canada
Contents Introduction to MTC Applications LTE Access Contribution System Model Simulation Results Conclusion
Introduction to MTC MTC is a fully automatic communication system between machine devices without necessary human intervention. A large number of terminals Small data transmissions. Low mobility Delay tolerance Spatially clustered 1/21
Applications (E-Healthcare (Body (Bridge (Smart (Power Area Monitoring) Cities) Network) Grid) System) 2/21
LTE Access 3/21
Reference Scheme LTE Network All machines directly access BS through physical random access channel (PRACH) Four message exchange 54 preambles 4/21
Random Access Channel (RACH) Model Machine randomly select a RACH BS Machine A Select the same RACH BS Collision Machine B Machine A Select different RACHs BS Success Machine B 6/21
PRACH Configuration Options 5/21
Contribution 7/21
Two-Hop Slotted ALOHA-based Clustering Scheme Isolated node Cluster member Cluster Head Resource allocation between slotted ALOHA (1 st hop) and PRACH (2 nd hop) e.g. 3:1 1 st hop 2 nd hop In each cluster, it fully reuse the resources for slotted ALOHA communication. Interference inside and outside the cluster. 8/21
Details of the Access Procedures ( Reference Scheme ) ( Proposed Scheme ) 9/21
Novelty Previous Research Our Research Intra-Cluster Communication Method Scheduled Scheme ( e.g. TDMA ) Slotted ALOHA Resources for Intra-Cluster Communication Unlicensed Resource (WiFi, ZigBee) Resources are migrated from PRACH, All-licensed spectrum Spatial Traffic Pattern Homogeneous (PPP) Clustered 10/21
System Model 11/21
Location Generator Internal Parameters: number of devices (N), number of clusters (M), radius of cluster (R), prob. of isolated nodes (P isolated ). 12/21
CoV-based Clustering Metrics Probabilistic Measure: Voronoi Tessellation areas Statistics: Voronoi scaled Tessellation: CoV Given a set of points (seed), the plane is partitioned into cells. Each cell corresponds to a seed, and this cell consists of all points that are closer to its seed than any others. σ is standard deviation, μ is the mean of Voronoi cell areas, k is the factor to normalize C V to 1 when the points are taken from a Poisson point process (PPP). M. Mirahsan, R. Schoenen, H. Yanikomeroglu, HetHetNets: Heterogeneous traffic distribution in heterogeneous wireless cellular networks, IEEE J. Sel. Areas Commun., vol. 33, no. 10, pp. 2252 2265, Oct. 2015. Voronoi of PPP traffic Voronoi of clustered traffic 13/21
Different Spatial Traffic and their C V 2000 points C V =1: Poisson Point Process. C V >1: super-poisson/cluster. 0<C V <1: sub-poisson. As traffic becomes more heterogeneous/clustered, C V becomes higher. 14/21
Results of Hierarchical Clustering Algorithm on Different Amounts of Clustering 2000 points CoV=6.2589 CoV=4.7055 CoV=3.4277 CoV=1.2936 15/21
Simulation Results 16/21
Simulation Parameters 3GPP LTE Network F. Alsewaidi, D. Kaleshi, A. Doufexi, Analysis of Radio Access Network Performance for M2M Communications in LTE-A at 800 MHz, in IEEE WCNC 2014 IoT Commun. & Techn. Workshop. 17/21
Reference Random Access Procedure 20
3GPP-Compliant LTE Random Access Simulator Performances Measures Number of MTC devices 5k 10k 30k Result Origin Collision Prob. (%) 0.01 0.03 0.22 3GPP 0.01 0.03 0.23 Simulation Success Prob. (%) 1.09 2.18 6.49 Simulation Idle Prob. (%) 98.90 97.79 0.93 Simulation Access Success Prob. (%) 100 100 100 3GPP 100 100 100 Simulation Avg. Access Delay (ms) Avg. Preamble Trans (%) 25.60 26.05 27.35 3GPP 28.23 28.58 29.63 Simulation 1.43 1.45 1.50 3GPP 1.43 1.45 1.50 Simulation RAN improvements for machine-type communications, 3rd Generation Partnership Project (3GPP), TR 37.868, Sept. 2012. 18/21
Performance vs. C V higher C V more clustered locations less Interference better performance Why such a big difference in energy consumption Fig. (d)? 19/21
Performance vs. Cluster No. (C v fixed) The other inputs of location generator are N = 2000 and P isolated = 0.01. More clusters Smaller cluster radius Less interference Better performance (C V fixed) The proposed scheme outperforms the reference scheme when 100 > M > 22, i.e., the number of machines per cluster < 90. 20/21
Conclusion Proposed a slotted ALOHA-based two-hop cluster random access method which significantly saves the energy for machines. Introduced a clustering geometry model for machine locations. Defined a clustering metric C V. Conducted some literature review on different clustering algorithms. Examined the impact of different parameters (C V, device no., Cluster no.) on the system performance. 21/21
25
Reference Random Access Procedure 26
Simulation System Flowchart Parameter setup and initialization Generate clustered points in space domain Cluster the MDs and classify the cluster nodes and isolated nodes Uniformly distribute access attempts over RA period Resource Allocation select CHs based on channel gains Run slotted ALOHA and RA procedure subframe by subframe until all devices complete their accesses Performance Analysis 27
System Model of Two-Hop Clustering Scheme a. Generate a traffic pattern b. Cluster all devices + select CHs c. Two-hop Communication: 1. Intra cluster Cluster members upload packets to cluster head through slotted Aloha 1. Direct link between CH/ individual node with BS Once CH buffer reaches a certain level, it initiates random access to BS. d. Resource allocation Resources for slotted Aloha are migrated from the original RACH to maintain the same amount of total resources and make a fair comparison. Full Resource Reuse for slotted aloha within each cluster. e. Interference Simultaneous packet transmission in slotted aloha will cause interference. Only if SINR is beyond a certain level, transmission is successful. f. Performance evaluation. Tuning different parameters to see the impact on performance. 28
Resource Allocation Scheme 1. Resource allocation between PRACH and slotted ALOHA e.g. 3:1 1 st hop 2 nd hop Note: More resources should be reserved for slotted ALOHA, Only cluster head and individual ME contend in PRACH. 2. Intra-cluster communication (slotted ALOHA) reuse pattern Full reuse in each cluster. Interference Any other simultaneous transmission from machines located at the same or different clusters will cause interference. If SINR>20dB, packet transmission is successful (slotted ALOHA). More clustered less inter-cluster interference. Channel Model ME transmit power 14dBm, Noise figure 9dB Pathloss exponent 4 (machine to machine) 29
Analysis of Energy Consumption wrt. Collision Probability In slotted ALOHA Transmission power: P Tx2 ; Receiving and processing power: P Rx2 ; Transmission period: T Tx ; Response window: T RESP ; Receiving period: T Rx ; The probability that a CM needs to transmit N times: The Energy consumed: The expected number of transmissions: The expected energy consumption: 30
Analysis of Energy Consumption wrt. Collision Probability In RACH Similarly, the expected energy consumption: 31
Analysis of Energy Consumption wrt. Collision Probability 32
Definition of Performance Metrics Total transmission times Reference Scheme: the total number of preambles sent by all machines in RACH Proposed Scheme: the total transmission times of all cluster members in slotted aloha + the total number of preambles sent by CH and individual node in RACH Average access delay Reference Scheme: from first access subframe to the access complete subframe Proposed Scheme slotted aloha: from the first packet transmission subframe to its packet successfully uploading subframe RACH: from first access subframe to the access complete subframe Average Energy Consumption Reference Scheme: energy dissipation in RACH Proposed Scheme: energy dissipation in slotted aloha+ RACH 33
Definition of Performance Metrics Collision rate: Reference scheme: only in RACH Cluster Scheme: both RACH and Slotted Aloha 34
Some Assumptions in Simulation For Slotted Aloha all packet transmission can be finished in one subframe. Use ACK to confirm successful transmission in slotted aloha the ACK response will be acknowledged immediately/in the current subframe. Packet can be decoded successfully only if Interference is under a certain level. For slotted aloha, backoff is done only across slotted aloha access slots Once uploading data, device state is complete with assumption his CH will get access successfully in future Buffer has no limit For Other Parts distribute access uniformly over RACH period 35